Dr. Bob Shillman
Analyst · Robert W. Baird. Please proceed
Okay. The general bin picking problem has been around since I was a graduate student 40 years ago. But at that time, it was presented as a tub of identical parts lying in any random array, and the goal was to find the one that was on top. And that pretty much is solvable. But the problems that customers want to solve today is quite a bit more difficult than that. The problem, as I understand the customers want to solve is there’s a bin, some place in a rack even, not necessarily on a laboratory floor; there is a bin in space and it contains numerous items not all the same matter of fact. Now if you ask a human, which into his bin without even looking, your hand over your head, and pick out the pen, the packets of the pen in it, instead of the packages that have calculators in them or whatever, a human can do that. And having been in this field for 40 years now, everyday I'm more and more impressed by what humans can do, even a low human. The ability to find things, to grasp them and if they grasp it, they grasp it appropriately and hard enough, so they won’t drop it even if they’re moving their hand quickly. So, the problems we’re facing today are that there’s a bin somewhere and there are mix parts in there in various orientation and the goal is to pick out the calculator out of that bin, which contains other things as well as calculators. Now the problem, even though humans can do it by a sense of touch and understanding what things feel like and how hard they are, how much they weigh, the problem requires today form automation to be solved, it does require machine vision to look into the beam, the first thing to do to look into the beam, try to find the part that's there and then more importantly, or as importantly, locate the point to pick it up. Because if a robot arm goes in there and picks it up from the end let's say, let's say it's a piece of wood measuring a foot long, if it picks it up by the end, by the time the hand moves back to put it in the shipping carton, it will have dropped that because how it turns out, the robots have to move quickly to justify this entire system. So, not only does the vision have to determine where the part is it is supposed to pick up, but also where to pick it up, then that's the vision problem very complex, very complex. Next, there is the gripper problem. What kind of gripper is going to be used to pick up the part. Well, there are a variety of different kinds of grippers, there are mechanical ones, suction ones, and how harder they squeeze on it, so that when they move quickly, they won't drop the part, very hard problem. Then, there is a robotic mechanics problem, which is how do you move the arm and the gripper in such a way to get into the bin, not knock the bin, not hit the other staging that is holding the bin and compute the path to pick it up and then to retrieve it and put it back in the box. So, this is a very, very difficult problem. I would say, it's almost equivalent to self-driving cars, which is again a very difficult problem. And although we have pieces of that puzzle, I don't believe that anyone yet has put the entire, all those pieces together into a solution. And we've been working with variety of companies to try to solve this problem. But I'm afraid that, it is beyond the technological state-of-the-art and will remain so for a few years.